* [PATCH] Microstate accounting: account for some more things
@ 2003-07-08 3:14 Peter Chubb
0 siblings, 0 replies; only message in thread
From: Peter Chubb @ 2003-07-08 3:14 UTC (permalink / raw)
To: linux-kernel
[-- Attachment #1: message body text --]
[-- Type: text/plain, Size: 476 bytes --]
Hi,
The appended patch goes on top of the patch at
http://marc.theaimsgroup.com/?l=linux-kernel&m=105662514529349&w=2
to add accounting for time spent in poll/select, sleeping handling
major pagefaults, and time spent asleep on futexes.
Running this patch, I found a problem with major fault accounting in
the kernel: there are combinations of circumstances that can lead to a
major fault being reported when no I/O is done (for example if the
page has been prefaulted)
[-- Attachment #2: p1 --]
[-- Type: text/plain, Size: 10614 bytes --]
# This is a BitKeeper generated patch for the following project:
# Project Name: Linux kernel tree
# This patch format is intended for GNU patch command version 2.5 or higher.
# This patch includes the following deltas:
# ChangeSet 1.1402 -> 1.1403
# fs/select.c 1.20 -> 1.21
# arch/i386/kernel/irq.c 1.39 -> 1.40
# kernel/futex.c 1.28 -> 1.29
# kernel/msa.c 1.1 -> 1.2
# arch/ia64/kernel/irq_ia64.c 1.14 -> 1.15
# mm/memory.c 1.125 -> 1.126
# fs/proc/base.c 1.52 -> 1.53
# include/linux/msa.h 1.1 -> 1.2
#
# The following is the BitKeeper ChangeSet Log
# --------------------------------------------
# 03/07/08 peterc@gelato.unsw.edu.au 1.1403
# Add new states: time sleeping for page I/O, in poll/select/epoll and in futex.
# --------------------------------------------
#
diff -Nru a/arch/i386/kernel/irq.c b/arch/i386/kernel/irq.c
--- a/arch/i386/kernel/irq.c Tue Jul 8 13:13:30 2003
+++ b/arch/i386/kernel/irq.c Tue Jul 8 13:13:30 2003
@@ -429,7 +429,7 @@
unsigned int status;
irq_enter();
- msa_start_irq(cpu, irq);
+ msa_start_irq(irq);
#ifdef CONFIG_DEBUG_STACKOVERFLOW
/* Debugging check for stack overflow: is there less than 1KB free? */
@@ -509,7 +509,7 @@
spin_unlock(&desc->lock);
irq_exit();
- msa_finish_irq(cpu, irq);
+ msa_finish_irq(irq);
return 1;
}
diff -Nru a/arch/ia64/kernel/irq_ia64.c b/arch/ia64/kernel/irq_ia64.c
--- a/arch/ia64/kernel/irq_ia64.c Tue Jul 8 13:13:30 2003
+++ b/arch/ia64/kernel/irq_ia64.c Tue Jul 8 13:13:30 2003
@@ -122,7 +122,7 @@
saved_tpr = ia64_get_tpr();
ia64_srlz_d();
oldvector = vector;
- msa_start_irq(cpu, local_vector_to_irq(vector));
+ msa_start_irq(local_vector_to_irq(vector));
while (vector != IA64_SPURIOUS_INT_VECTOR) {
if (!IS_RESCHEDULE(vector)) {
ia64_set_tpr(vector);
@@ -139,7 +139,7 @@
ia64_eoi();
oldvector = vector;
vector = ia64_get_ivr();
- msa_continue_irq(cpu, local_vector_to_irq(oldvector),
+ msa_continue_irq(local_vector_to_irq(oldvector),
local_vector_to_irq(vector));
}
/*
@@ -150,7 +150,7 @@
if (local_softirq_pending())
do_softirq();
- msa_finish_irq(cpu, local_vector_to_irq(vector));
+ msa_finish_irq(local_vector_to_irq(vector));
}
#ifdef CONFIG_SMP
diff -Nru a/fs/proc/base.c b/fs/proc/base.c
--- a/fs/proc/base.c Tue Jul 8 13:13:30 2003
+++ b/fs/proc/base.c Tue Jul 8 13:13:30 2003
@@ -302,9 +302,7 @@
static int proc_pid_msa(struct task_struct *task, char *buffer)
{
struct microstates *msp = &task->microstates;
-
static char *statenames[] = {
- "Unknown",
"User",
"System",
"Interruptible",
@@ -313,32 +311,40 @@
"OnExpiredQueue",
"Zombie",
"Stopped",
+ "Paging",
+ "Futex",
+ "Poll",
"Interrupted",
};
+
return sprintf(buffer,
- "State: %s\n" \
- "ONCPU_USER %15llu\n" \
- "ONCPU_SYS %15llu\n" \
- "INTERRUPTIBLE %15llu\n" \
- "UNINTERRUPTIBLE%15llu\n" \
- "INTERRUPTED %15llu\n" \
- "ACTIVEQUEUE %15llu\n" \
- "EXPIREDQUEUE %15llu\n" \
- "STOPPED %15llu\n" \
- "ZOMBIE %15llu\n" \
- "UNKNOWN %15llu\n",
- msp->cur_state <= INTERRUPTED ?
+ "State: %s\n" \
+ "ONCPU_USER %15llu\n" \
+ "ONCPU_SYS %15llu\n" \
+ "INTERRUPTIBLE %15llu\n" \
+ "UNINTERRUPTIBLE%15llu\n" \
+ "INTERRUPTED %15llu\n" \
+ "ACTIVEQUEUE %15llu\n" \
+ "EXPIREDQUEUE %15llu\n" \
+ "STOPPED %15llu\n" \
+ "ZOMBIE %15llu\n" \
+ "SLP_POLL %15llu\n" \
+ "SLP_PAGING %15llu\n" \
+ "SLP_FUTEX %15llu\n", \
+ msp->cur_state >= 0 && msp->cur_state < NR_MICRO_STATES ?
statenames[msp->cur_state] : "Impossible",
- msp->timers[ONCPU_USER],
- msp->timers[ONCPU_SYS],
- msp->timers[INTERRUPTIBLE_SLEEP],
- msp->timers[UNINTERRUPTIBLE_SLEEP],
- msp->timers[INTERRUPTED],
- msp->timers[ONACTIVEQUEUE],
- msp->timers[ONEXPIREDQUEUE],
- msp->timers[STOPPED],
- msp->timers[ZOMBIE],
- msp->timers[UNKNOWN]);
+ msp->timers[ONCPU_USER],
+ msp->timers[ONCPU_SYS],
+ msp->timers[INTERRUPTIBLE_SLEEP],
+ msp->timers[UNINTERRUPTIBLE_SLEEP],
+ msp->timers[INTERRUPTED],
+ msp->timers[ONACTIVEQUEUE],
+ msp->timers[ONEXPIREDQUEUE],
+ msp->timers[STOPPED],
+ msp->timers[ZOMBIE],
+ msp->timers[POLL_SLEEP],
+ msp->timers[PAGING_SLEEP],
+ msp->timers[FUTEX_SLEEP]);
}
#endif /* CONFIG_MICROSTATE */
diff -Nru a/fs/select.c b/fs/select.c
--- a/fs/select.c Tue Jul 8 13:13:30 2003
+++ b/fs/select.c Tue Jul 8 13:13:30 2003
@@ -253,6 +253,7 @@
retval = table.error;
break;
}
+ msa_next_state(current, POLL_SLEEP);
__timeout = schedule_timeout(__timeout);
}
__set_current_state(TASK_RUNNING);
@@ -443,6 +444,7 @@
count = wait->error;
if (count)
break;
+ msa_next_state(current, POLL_SLEEP);
timeout = schedule_timeout(timeout);
}
__set_current_state(TASK_RUNNING);
diff -Nru a/include/linux/msa.h b/include/linux/msa.h
--- a/include/linux/msa.h Tue Jul 8 13:13:30 2003
+++ b/include/linux/msa.h Tue Jul 8 13:13:30 2003
@@ -18,7 +18,7 @@
*/
enum thread_state {
- UNKNOWN,
+ UNKNOWN = -1,
ONCPU_USER,
ONCPU_SYS,
INTERRUPTIBLE_SLEEP,
@@ -28,6 +28,10 @@
ZOMBIE,
STOPPED,
INTERRUPTED,
+ PAGING_SLEEP,
+ FUTEX_SLEEP,
+ POLL_SLEEP,
+
NR_MICRO_STATES /* Must be last */
};
@@ -76,9 +80,9 @@
void msa_update_parent(struct task_struct *parent, struct task_struct *this);
void msa_init(struct task_struct *p);
void msa_set_timer(struct task_struct *p, int state);
-void msa_start_irq(int cpu, int irq);
-void msa_continue_irq(int cpu, int oldirq, int newirq);
-void msa_finish_irq(int cpu, int irq);
+void msa_start_irq(int irq);
+void msa_continue_irq(int oldirq, int newirq);
+void msa_finish_irq(int irq);
clk_t msa_irq_time(int cpu, int irq);
#ifdef TASK_STRUCT_DEFINED
diff -Nru a/kernel/futex.c b/kernel/futex.c
--- a/kernel/futex.c Tue Jul 8 13:13:30 2003
+++ b/kernel/futex.c Tue Jul 8 13:13:30 2003
@@ -34,6 +34,7 @@
#include <linux/futex.h>
#include <linux/vcache.h>
#include <linux/mount.h>
+#include <linux/msa.h>
#define FUTEX_HASHBITS 8
@@ -349,6 +350,7 @@
* the waiter from the list.
*/
add_wait_queue(&q.waiters, &wait);
+ msa_next_state(current, FUTEX_SLEEP);
set_current_state(TASK_INTERRUPTIBLE);
if (!list_empty(&q.list)) {
unlock_futex_mm();
diff -Nru a/kernel/msa.c b/kernel/msa.c
--- a/kernel/msa.c Tue Jul 8 13:13:30 2003
+++ b/kernel/msa.c Tue Jul 8 13:13:30 2003
@@ -62,36 +62,78 @@
msnext->timers[msnext->cur_state] += now - msnext->last_change;
/* Update states */
- if (msprev->next_state != UNKNOWN) {
- next_state = msprev->next_state;
- msprev->next_state = UNKNOWN;
- } else switch(prev->state) {
- case TASK_INTERRUPTIBLE:
- next_state = INTERRUPTIBLE_SLEEP;
- break;
+ switch (msprev->next_state) {
+ case UNKNOWN: /*
+ * Infer from actual state
+ */
+ switch (prev->state) {
+ case TASK_INTERRUPTIBLE:
+ next_state = INTERRUPTIBLE_SLEEP;
+ break;
- case TASK_UNINTERRUPTIBLE:
- next_state = UNINTERRUPTIBLE_SLEEP;
- break;
-
- case TASK_STOPPED:
- next_state = STOPPED;
- break;
-
- case TASK_ZOMBIE:
- next_state = ZOMBIE;
- break;
-
- case TASK_DEAD:
- next_state = ZOMBIE;
+ case TASK_UNINTERRUPTIBLE:
+ next_state = UNINTERRUPTIBLE_SLEEP;
+ break;
+
+ case TASK_STOPPED:
+ next_state = STOPPED;
+ break;
+
+ case TASK_ZOMBIE:
+ next_state = ZOMBIE;
+ break;
+
+ case TASK_DEAD:
+ next_state = ZOMBIE;
+ break;
+
+ case TASK_RUNNING:
+ next_state = ONACTIVEQUEUE;
+ break;
+
+ default:
+ next_state = UNKNOWN;
+ break;
+
+ }
+ break;
+
+ case PAGING_SLEEP: /*
+ * Sleep states are PAGING_SLEEP;
+ * others inferred from task state
+ */
+ switch(prev->state) {
+ case TASK_INTERRUPTIBLE: /* FALLTHROUGH */
+ case TASK_UNINTERRUPTIBLE:
+ next_state = PAGING_SLEEP;
+ break;
+
+ case TASK_STOPPED:
+ next_state = STOPPED;
+ break;
+
+ case TASK_ZOMBIE:
+ next_state = ZOMBIE;
+ break;
+
+ case TASK_DEAD:
+ next_state = ZOMBIE;
+ break;
+
+ case TASK_RUNNING:
+ next_state = ONACTIVEQUEUE;
+ break;
+
+ default:
+ next_state = UNKNOWN;
+ break;
+ }
break;
- case TASK_RUNNING:
- next_state = ONACTIVEQUEUE;
+ default: /* Explicitly set next state */
+ next_state = msprev->next_state;
+ msprev->next_state = UNKNOWN;
break;
-
- default:
- next_state = UNKNOWN;
}
msprev->cur_state = next_state;
@@ -159,11 +201,12 @@
}
}
-void msa_start_irq(int cpu, int irq)
+void msa_start_irq(int irq)
{
struct task_struct *p = current;
struct microstates *mp = &p->microstates;
clk_t now;
+ int cpu = smp_processor_id();
MSA_NOW(now);
mp->timers[mp->cur_state] += now - mp->last_change;
@@ -175,9 +218,10 @@
msa_irq_pids[cpu][irq] = current->pid;
}
-void msa_continue_irq(int cpu, int oldirq, int newirq)
+void msa_continue_irq(int oldirq, int newirq)
{
clk_t now;
+ int cpu = smp_processor_id();
MSA_NOW(now);
msa_irq_times[cpu][oldirq] += now - msa_irq_entered[cpu][oldirq];
@@ -186,11 +230,12 @@
}
-void msa_finish_irq(int cpu, int irq)
+void msa_finish_irq(int irq)
{
struct task_struct *p = current;
struct microstates *mp = &p->microstates;
clk_t now;
+ int cpu = smp_processor_id();
MSA_NOW(now);
@@ -249,7 +294,7 @@
}
tp = which == MSA_SELF ? msp->timers : msp->child_timers;
- for (i = UNKNOWN; i < ntimers; i++) {
+ for (i = 0; i < ntimers; i++) {
clk_t x = MSA_TO_NSEC(*tp++);
if (copy_to_user(timers++, &x, sizeof x))
return -EFAULT;
diff -Nru a/mm/memory.c b/mm/memory.c
--- a/mm/memory.c Tue Jul 8 13:13:30 2003
+++ b/mm/memory.c Tue Jul 8 13:13:30 2003
@@ -1532,6 +1532,7 @@
if (is_vm_hugetlb_page(vma))
return VM_FAULT_SIGBUS; /* mapping truncation does this. */
+ msa_next_state(current, PAGING_SLEEP);
/*
* We need the page table lock to synchronize with kswapd
* and the SMP-safe atomic PTE updates.
@@ -1541,10 +1542,14 @@
if (pmd) {
pte_t * pte = pte_alloc_map(mm, pmd, address);
- if (pte)
- return handle_pte_fault(mm, vma, address, write_access, pte, pmd);
+ if (pte) {
+ int ret = handle_pte_fault(mm, vma, address, write_access, pte, pmd);
+ msa_next_state(current, UNKNOWN);
+ return ret;
+ }
}
spin_unlock(&mm->page_table_lock);
+ msa_next_state(current, UNKNOWN);
return VM_FAULT_OOM;
}
^ permalink raw reply [flat|nested] only message in thread
only message in thread, other threads:[~2003-07-08 3:00 UTC | newest]
Thread overview: (only message) (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2003-07-08 3:14 [PATCH] Microstate accounting: account for some more things Peter Chubb
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).